Circulating Re-Entrant Waves Promote Maturation of Hipsc-Derived Cardiomyocytes in Self-Organized Tissue Ring

ARTICLE

https://doi.org/10.1038/s42003-020-0853-0 OPEN Circulating re-entrant waves promote maturation of hiPSC-derived cardiomyocytes in self-organized tissue ring

Junjun Li 1,11, Lu Zhang2,11, Leqian Yu3, Itsunari Minami4, Shigeru Miyagawa1, Marcel Hörning 3,10, Ji Dong5,

1234567890():,; Jing Qiao3, Xiang Qu1, Ying Hua1, Nanae Fujimoto1, Yuji Shiba6, Yang Zhao7, Fuchou Tang5, Yong Chen3,8, ✉ ✉ ✉ Yoshiki Sawa1 , Chao Tang 2 & Li Liu1,9

Directed differentiation methods allow acquisition of high-purity cardiomyocytes differentiated from human induced pluripotent stem cells (hiPSCs); however, their immaturity characteristic limits their application for drug screening and regenerative therapy. The rapid electrical pacing of cardiomyocytes has been used for efficiently promoting the maturation of cardiomyocytes, here we describe a simple device in modified culture plate on which hiPSC- derived cardiomyocytes can form three-dimensional self-organized tissue rings (SOTRs). Using calcium imaging, we show that within the ring, reentrant waves (ReWs) of action potential spontaneously originated and ran robustly at a frequency up to 4 Hz. After 2 weeks, SOTRs with ReWs show higher maturation including structural organization, increased cardiac-specific gene expression, enhanced Ca2+-handling properties, an increased oxygen- consumption rate, and enhanced contractile force. We subsequently use a mathematical model to interpret the origination, propagation, and long-term behavior of the ReWs within the SOTRs.

1 Department of Cardiovascular Surgery, Osaka University Graduate School of Medicine, 2-2 Yamadaoka, Suita, Osaka 565-0871, Japan. 2 Center for Quantitative Biology and Peking-Tsinghua Center for Life Sciences, Academy for Advanced Interdisciplinary Studies, Peking University, 100871 Beijing, China. 3 Institutes for Integrated Cell-Material Sciences (WPI-iCeMS), Kyoto University, Yoshida-Ushinomiya-cho, Sakyo-ku, Kyoto 606-8501, Japan. 4 Department of Cell Design for Tissue Construction Faculty of Medicine, Osaka University, Osaka 565-0871, Japan. 5 Biomedical Pioneering Innovation Center, College of Life Sciences, Peking University, 100871 Beijing, China. 6 Department of Regenerative Science and Medicine, Institute for Biomedical Sciences, Shinshu University, 3-1-1 Asahi, Matsumoto, Nagano 390-0821, Japan. 7 State Key Laboratory of Natural and Biomimetic Drugs, The MOE Key Laboratory of Cell Proliferation and Differentiation, Institute of Molecular Medicine, Peking-Tsinghua Center for Life Sciences, Peking University, 100871 Beijing, China. 8 PASTEUR, Département de chimie, école normale supérieure, PSL Research University, Sorbonne Universités, UPMC Université Paris 06, CNRS, Paris 75005, France. 9 Department of Drug Discovery Cardiovascular Regeneration, Osaka University Graduate School of Medicine, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan. 10Present address: Institute of Biomaterials and Biomolecular Systems, University of Stuttgart, 70569 Stuttgart, Germany. 11These authors ✉ contributed equally: Junjun Li, Lu Zhang. email: [email protected]; [email protected]; [email protected]

COMMUNICATIONS BIOLOGY | (2020) 3:122 | https://doi.org/10.1038/s42003-020-0853-0 | www.nature.com/commsbio 1 ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-0853-0

lthough human-induced pluripotent stem cell (hiPSC) cardiomyocytes aggregated and formed a thick tissue ring within Aderived cardiomyocytes have been proposed as an abun- 2 days (432.72 ± 56.18 µm on day 2, Fig. 1a–c and Supplementary dant resource for tissue engineering, drug screening, and Fig. 1). As indicated by the genetically encoded calcium indicator regenerative-medicine applications1, they exhibit characteristics GCaMP3, we found looped activation propagations within the different from adult human cardiomyocytes, including immature SOTRs (i.e., ReWs; Supplementary Movie 1), with zero to two sarcomere structure and morphology, a fetus-like gene expression ReWs present in one SOTR (Fig. 1d, e and Supplementary profile, and inadequate Ca2+-handling properties2–4. The Movie 2). Besides circular geometry, other geometries have also immature nature of hiPSC-derived cardiomyocytes potentially been tested, such as 4-point star, 5-point star (Supplementary hinders their reflection of adult heart physiology for disease Fig. 2a), and the geometry between rectangular and circular could modeling and drug assessment. also be obtained. However, the formation of the tissues To achieve higher degrees of hiPSC-derived cardiomyocytes throughout different areas of these templates was less homo- maturation and function, a variety of methods have been devel- geneous than that of the tissues in the circular geometry. In oped, including dynamic culture5, three-dimensional (3D) engi- addition, the circular ring tissues were more equally and stably neered heart tissue6–8, 3D printing9, addition of factors10, organized while maintaining the ReWs with a higher ratio extracellular matrix10,11, and external stimulation. Electrical sti- (Supplementary Fig. 2b). Thus, we chose the circular ring to mulation, especially high-frequency pacing, has long been sug- engineer cardiac tissues. Fluorescence-activated cell sorting gested as an effective method for maturing muscle cells3,12–15, (FACS) (Supplementary Fig. 1b), voltage dye staining, and which upregulates cardiac markers13,14,16, enhances Ca2+-hand- immunostaining data indicated that the SOTRs were mostly ling properties3, and promotes cardiomyocyte alignment17. ventricular cardiomyocytes and few fibroblast cells (Supplemen- During this process, appropriate stimulation protocols are tary Figs. 3 and 4). required to minimize possible tissue damage18. Additionally, it A trace recording (Fig. 1d, e) indicated no long rest periods remains challenging to rapidly pace cardiomyocyte tissues (≥2 between successive ReWs, resulting in higher beating rates in Hz) over long time19 due to possible side effects, such as heavy SOTRs capable of sustaining a higher number of ReWs (Fig. 1e, metal poisoning, electrolysis, pH shift, and the generation of f). The spontaneous beating rate of SOTRs with zero ReW was reactive oxygen species (ROS)20,21. Moreover, upscaling for mass 0.21 ± 0.10 Hz on day 6 and remained stable for 2 weeks, whereas stimulation is either difficult or requires high level of power the rates of SOTRs with one or two ReWs on day 6 were much consumption19. As an important complement, mechanical sti- higher (2.40 ± 0.49 and 3.30 ± 0.39 Hz, respectively). Notably, the mulation can promote cardiomyocyte maturation, with cyclic wave speed in the ReW groups was much lower (~2 cm s−1) mechanical stress applied by external stretching devices capable than that of the spontaneous beating group (0 ReW; 5.93 ± of forcing muscle-cell assembly into structurally and functionally 1.50 cm s−1). It is possible that the slower speed together with the aligned 3D syncytium22, thereby affecting their gene expres- shortened pacing interval and refractory period might have been sion23–25 and Ca2+ cycling26. caused by the higher beating frequency in the ReW groups On the other hand, spiral waves [re-entrant waves (ReWs)] relative to that in the zero ReW group (Supplementary Fig. 5 and within cardiac tissue have long been investigated as a model for Supplementary Movie 3), which agrees with previous reports arrhythmia studies27–31, and the rapid beating caused by in vivo associated with excitable media32–36. Additionally, after the ReW spiral wave could lead to dysfunction of heart and even death of a was stopped, the speed of spontaneous beating in the two ReW patient. However, when used in an in vitro condition, the rapid group increased significantly (p = 0.01) to more than 10 cm s−1 beating caused by spiral wave, similar to rapid electrical stimu- (Fig. 1g). lation14, could be beneficial for cardiomyocytes’ maturation. ThebeatingratesoftheoneandtwoReWgroupsincreased In this study, we create a platform capable of promoting rapid slightly over 2 weeks of culturing, reaching 3.03 ± 0.68 and 3.92 formation of hiPSC-derived cardiomyocytes into 3D self- ± 0.69 Hz on day 14, respectively. After a 6-day culture period, organized tissue rings (SOTRs), where propagation of an action 83.8% of 204 SOTRs had one or two ReWs, 14.2% had zero potential in the form of ReWs can spontaneously originate and ReW, and 2% had three ReWs; however, these percentages travel around the closed-loop circuit, thereby making the cardi- changed after 2 weeks of culture to 28.5% with zero ReW, 48.9% omyocytes beat at a high frequency (~2–4 Hz) continuously and with one ReW, 22.6% with two ReWs, and no SOTRs with three robustly up to more than 89 days without any external stimula- ReWs (Fig. 1h). Noise and disturbances, such as those during tion. Additionally, we found that the beating frequency and the medium changing, might have provoked changes in and/or ReW speed can be adjusted by changing the diameter of the ring. disappearance of ReWs. In the long-term culture, we found that Furthermore, we construct a mathematical model in order to ReWs could be maintained in SOTRs for >89 days (Supple- elucidate the origination, propagation, and long-term behavior of mentary Fig. 6). In addition to the static culture, dynamic the ReWs, with the model ultimately agreeing well with the culture of SOTRs on a rotary shaker has been applied to check experimental data. After 2 weeks of culture, the SOTRs with the effect on ReWs. During the culture, we did find the ReWs demonstrate improved structural organization, upregulated improvement of beating frequency of cardiomyocytes in the cardiac-specific gene expression, enhanced Ca2+-handling prop- ring with ReWs (Supplementary Fig. 7), which significantly erties, an increased oxygen-consumption rate (OCR), and increased from 3.30 ± 0.39 to 3.89 ± 0.18 Hz at day 6 (p = 0.038) enhanced contractile force. Our results demonstrate, to our and from 3.92 ± 0.69 to 5.57 ± 0.06 Hz at day 14 (p = 0.001). knowledge, a novel approach for spontaneous cardiomyocyte The higher beating frequency has been previously reported to maturation and could serve as an economical and practical sys- be associated with higher maturation of cardiomyocytes14.On tem for future production of mature hiPSC-derived another hand, there are no significant differences between the cardiomyocytes. zero ReW groups with or without rotatory culture. These data indicated dynamic culture could improve the availability of oxygen and glucose to cardiomyocytes in the ReW group5. Results Despite the improvement, we also found that the ReW Self-organization of a hiPSC-derived cardiomyocyte ring.We occurrence ratio in the dynamic group dropped to 50% on created 3D SOTRs by plating hiPSC-derived cardiomyocytes in a day 6 and 40% on day 14; the disappearance of the ReW was culture dish with a pillar in the center, around which the perhaps due to the disturbance caused by the dynamic medium

2 COMMUNICATIONS BIOLOGY | (2020) 3:122 | https://doi.org/10.1038/s42003-020-0853-0 | www.nature.com/commsbio COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-0853-0 ARTICLE

a c 3 PDMS Day 0 Day 1 Day 2 m) 2 m ** 1 ** ** ** Low adhesion substrate Cardiomyocytes Self-Organized Tissue Ring (SOTR) Width ( 0 02468101214 Culture time (days)

Day 0 Day 1 Day 2 Day 6

0ReW 1ReW 2ReWs f 5 ** ** de ** ** ** ** 0ReW 0ReW 4 ** ** ** * ** 2 peaks

a.u.) 3 ** (

1 Intensity Beat frequency (Hz) 0936 0 2 6 10 14 Culture time (days) 1ReW 1ReW g

.) p= ) * 0.01 1 - .u 21 peaks *p=0.045

a 15 ms ty ( c i d( ns 10 e e p Inte

es 5

0936 v

Wa 0 2 ReWs 2 ReWs 0ReW 1ReW 2ReWs 29 peaks h Time (ms) 0ReW 1ReW 2ReWs 3ReWs 780 100 585 390

195 tio (%) 75 a

0 Intensity (a.u.) 0936 er 50 nc Time (s) 25 Occurre 0 Day 6 Day 14

Fig. 1 ReWs promoted rapid beating of cells in SOTRs. a Schematic describing SOTR formation. b Bright-field images of hiPSC-derived cardiomyocytes in the template. The red arrows indicate the edge of cardiac tissues in the template, and the dashed lines indicate the PDMS block and pillar boundaries, respectively. Pillar diameter = 3 mm. Scale bar represents 800 µm. c Quantification of the width of the SOTRs on the indicated culture day (mean ± s.d.; n = 4 biologically independent samples). **P < 0.001 (ANOVA). d Activation map of GCaMP3-positive SOTRs with zero, one, or two ReWs. The red arrows indicate the propagation direction of the action potential. e GCaMP3 fluorescence signal at a fixed position on the ring of SOTRs with zero, one, or two ReWs on day 6. f Beat rates of SOTRs at different culture times (mean ± s.d.; 2 ReWs: n = 10; 1 ReW: n = 12; 0 ReW: n = 8 biologically independent samples). *P < 0.05; **P < 0.001 (ANOVA). g The wave speed of spontaneous beating in SOTRs with zero, one, or two ReWs after 14 days of culture; for ReW groups, the speed was recorded after the ReWs are stopped and the spontaneous beating was recovered (mean ± s.d.; 2 ReWs: n = 4; 1 ReW: n = 4; 0 ReW: n = 6 biologically independent samples). *P < 0.05 (ANOVA). h The percentage of occurrence for different numbers of ReWs on days 6 (n = 204 biologically independent samples) and 14 (n = 186 biologically independent samples), respectively.

COMMUNICATIONS BIOLOGY | (2020) 3:122 | https://doi.org/10.1038/s42003-020-0853-0 | www.nature.com/commsbio 3 ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-0853-0

flow. More future research work is necessary to improve the beat at various frequencies in the absence of external culture well design and reduce the disturbance of medium. stimulation. To better understand the nature of these ReWs according to We used RNA-sequencing to compare the gene expression their organization, propagation, and long time behavior, we profiles among different groups. The principal component constructed a mathematical model (Supplementary Information analysis (PCA; Fig. 3a) and hierarchical clustering of Spearman’s and Supplementary Fig. 8) comprising a ring of cells, each of correlation results (Fig. 3b) showed closer correlations between which can beat spontaneously with an intrinsic frequency. The hiPSC-derived cardiomyocytes receiving training by ReWs as total number of the cells in each ring in the model is proportional compared with those without ReW training. Additionally, we to the diameter of the ring. The coupling between neighboring found different gene expression patterns among zero, one, and cells through gap junctions37,38 gradually increases with time in two ReW groups (Fig. 3c). Functional annotation according to order to simulate the self-organization and formation processes of Gene Ontology (GO) analysis revealed that the upregulated genes the SOTR. Initially, all cells were independently set to a random (adjusted P < 0.05; fold change >1.5) in SOTRs with two ReWs, phase of beating, with all cells beating independently. As the gap compared to those with no ReWs, were related to the response to junctions form and strengthen, the beating of one cell harbors an unfolded protein and a number of maturation-related terms. increasing probability of triggering beating in its neighboring cell, Most of the GO-enriched terms were associated with cardiac thereby forming a propagating wave. Initially, these waves are development, including muscle-structure development, unsynchronized and short lived, as they initiate, disappear, and extracellular-structure organization, actin-filament-based pro- sometimes meet and collide (Supplementary Movie 4). Over time, cesses, tissue morphogenesis, and muscle-system processes. The a SOTR is left with one dominant propagating-wave mode response to unfolded protein is a coping response to mitigate or comprising zero or more ReWs (Fig. 2a and Supplementary eliminate endoplasmic reticulum (ER) stress caused by hypoxia40, Movie 4), a process similar to that observed in our experimental which could result from the high-frequency beating of the SOTRs findings (Supplementary Movie 5). Notably, SOTRs with one or with ReWs. To verify these findings, we performed immunos- two ReWs accounted for most of the simulation samples, a trend taining and quantitative polymerase chain reaction (qPCR) also observed in our experimental results (Fig. 2b). analysis. Immunostaining for cardiac-specific markers clearly We then investigated how the ReW features and properties revealed higher expression of β-myosin heavy chain (β-MHC), a were affected by characteristics, such as ring diameter, in the cardiac maturity marker correlated with contractile velocity, mathematical model and experiments. The beating frequency of within SOTRs with one and two ReWs as compared with levels cells in SOTRs decreased with increasing ring diameter, as shown observed in the zero ReW group (Fig. 3e, f). Moreover, analysis of with one ReW in Fig. 2c. We noticed that a constant ReW speed messenger RNA (mRNA) expression revealed a number of implied a linear decrease in beating frequency with ring diameter upregulated genes, including MYH7 (encoding β-MHC) and (which is proportional to the ring perimeter), and that the greater genes involved in sarcomere structures (ACTN2 and DES), the perimeter, the more time the wave would spend to travel ventricular structures (MYL2 and MYL3), ER-Ca2+ function around it. However, the data in Fig. 2c show a slower-than-linear (PLN), myoglobin (MB), and β1-adrenoceptor (ADRB1). SOTRs decrease; therefore, we measured the wave speed, finding that it with ReWs had higher-fold upregulated genes as compared with increased along the ring diameter in both the simulation and those with zero ReW (Fig. 3e–g) Owing to the higher beating experiment (Fig. 2d). As discussed in the Supplementary rates during culture, which agreed well with previous reports Information, this was due to the ability of a single hiPSC- utilizing electrical stimulation at different frequencies3,16. derived cardiomyocyte to spontaneously beat. Cells in larger rings When the RNA-sequencing data of SOTR cardiomyocytes was would wait longer for the next ReW to arrive, thereby making compared with those of adult heart cells and hiPSC-derived them easier to be activated by the wave front and resulting in a cardiomyocytes from a previous report41 (Supplementary Fig. 10), faster wave speed. This insight allowed us to predict the speed of the SOTR cardiomyocytes showed lower nodal- (TBX18, SHOX2, two and three ReWs. As shown in Fig. 2e, the prediction agreed MSX2, TBX2, HCN1, HCN4) and atrial-related (MYH6, NPPA) well with our experimental results. Additionally, the mathema- and higher ventricular-related (MYL2) gene expression than the tical model showed that the maximum number of ReWs that a previously reported cardiomyocytes41. Both the SOTR cardio- SOTR could sustain increased along with ring diameter, which myocytes and cardiomyocytes of the previous study showed lower was confirmed experimentally (Fig. 2f). Here, the mathematical expression of MYH7 and MYL2, but higher IRX4 expression than model predicted that SOTRs with a diameter of 17 mm could the adult cardiomyocytes. In addition, after ReW training, the gap contain as many as 19 ReWs. Experimentally, we found that junction (GJA1) and the n-cadherin (CDH2) in two ReW group SOTRs with a 3-mm pillar were optimal based on their highest was higher than that in zero ReW group (Supplementary Figs. 11 occurrence of ReWs (~90% at ≥1 ReW) (Supplementary Fig. 9), and 12). which was probably caused by the passive stretch in the SOTRs Hematoxylin and eosin (HE) staining showed cardiomyocytes with a different pillar diameter and its effect on the anisotropic with ReWs densely packed with each other to a greater degree structural organization. Moreover, the beating frequency of one than those without ReWs (Supplementary Fig. 13). Moreover, ReW was only slightly slower than that in SOTRs with a 1-mm SOTRs with ReWs demonstrated densely packed cardiac pillar; therefore, SOTRs with a 3-mm pillar were chosen for myofilaments along the ring orientation, which was also along subsequent investigations. the path of the ReWs, whereas cardiomyocytes in SOTRs without ReWs (0 ReW) were randomly oriented and poorly organized (Fig. 4a, b). Additionally, after a 14-day culture period, ReWs are associated with maturation of cardiomyocytes.The cardiomyocytes in SOTRs with ReWs strongly expressed the α- human fetal heart rate varies, although it generally stabilizes at actinin as compared with those in SOTRs without ReWs (Fig. 4c), ~3 Hz, whereas the adult human heart rate is ~1 Hz39.Elec- which agrees with the gene expression results (ACTN2, Fig. 3g). trical stimulation has been used to mature cardiomyocytes Importantly, cardiomyocytes in SOTRs with ReWs exhibited by pacing their beating at a certain higher-than-normal fre- significantly longer sarcomere length than those without ReWs (1 quency relative to a normal human rate3; however, continuous ReW: 1.71 ± 0.08 µm; 2 ReWs: 1.83 ± 0.10 µm; and 0 ReW: 1.49 ± and high electrical stimulation can cause cell damage. Here, 0.08 µm; 1 ReW vs. 0 ReW: p = 0.012; 2 ReWs vs. 0 ReW: p = ReWs within the SOTRs were able to make the cardiomyocytes 0.001) (Fig. 4c, d), which is closer to the sarcomere length in

4 COMMUNICATIONS BIOLOGY | (2020) 3:122 | https://doi.org/10.1038/s42003-020-0853-0 | www.nature.com/commsbio COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-0853-0 ARTICLE

p 20 0ReW 1ReW 2ReWs 3ReWs

-20

-40

-60

-80 mV

0ReW 1ReW 2ReWs 3ReWs 800

600

400

200 Activation map Membrane potential ma 0 ms

bcd 60% 4 4.0 Simulation Simulation Experiment Simulation Experiment 3.5 ** 3.5 50% Experiment ** )

** -1 ** 3 s 3.0 **

40% m 2.5 2.5

30% 2 2.0 SOTRs

1.5 espeed(c 1.5 20% v Wa

Beat frequency (Hz) 1 1.0 10% Percentage of RWs in 3mm 0.5 0.5

0% 0 0.0 0ReW 1ReW 2ReWs 3ReWs 135 135 Diameter of pillar (mm) Diameter of pillar (mm) e f 4 25 Experiment Simulation Experiment Simulation ** 20 ) 1

- Calculation 3 ** ReWs * f o

(cm s 15 er b

2 m eed u n 10 m u

1 m Wave sp 5 Maxi

0 0 1ReW 2ReWs 3ReWs 1 3 5 7 9 11131517 The number of ReWs Diameter of pillar (mm)

Fig. 2 Accurate reproduction of the ReW features in SOTRs using a mathematical model. a Examples of simulation results of the model (upper panel) and the activation map (lower panel). For the group with zero ReW, an action potential initiates from the left position of the ring and propagates in the opposite directions along the ring, after which the two waves meet and annihilate on another. The process starts again with new waves initiating at the same position periodically. For groups with one, two, or three ReWs, stable ReWs travel around the ring. Membrane potential and activation time are color coded. b The percentage of stable ReWs in SOTRs at day 6 (diameter = 3 mm; n = 60 in the simulation and n = 204 biologically independent samples in the experiment). c Beat rate (Hz) and d ReW speed in SOTRs with one ReW at day 6 for different pillar diameters [experiment: 1-mm SOTR (n = 4); 3-mm SOTR (n = 12); 5-mm SOTR (n = 10 biologically independent samples)]. **P < 0.001 (ANOVA). e Speeds of one, two, or three ReWs in SOTRs with a pillar diameter of 3 mm [experiment: mean ± s.d.; 3 ReWs (n = 4); 2 ReWs (n = 10); 1 ReW (n = 12 biologically independent samples)]. *P < 0.05; **P < 0.001 (ANOVA). f The maximum number of ReWs in SOTRs with different pillar diameters. human adult cardiomyocytes42. Moreover, on day 2 the SOTRs ReWs compared to those with zero ReW. Our findings confirmed with or without ReWs all demonstrated randomly distributed and the genetic and structural maturation of SOTRs with ReWs. poorly organized cardiomyocytes, and similar maturation level (Supplementary Fig. 4). ReW could also be induced in zero ReW SOTR by rapid electrical stimulation at day 14 (Supplementary ReWs are associated with improved Ca2+-handling properties. Movie 6). Collectively, these data, together with the previous day An extracellular flux analyzer was used to characterize mito- 14 results, indicate that that the ReW pacing are associated with chondrial function in cardiomyocytes within SOTRs. To record improved maturation. the maximum activity of the electron-transport chain from ade- Furthermore, electron microscopy indicated that cells in the nosine triphosphate (ATP) synthesis, mitochondrial ATP syn- ReW groups exhibited larger sarcomeric bundles, well-defined Z thase was inhibited by oligomycin, and a proton-gradient disks, A-bands, and myofibrils (Fig. 4e). These data suggest that discharger was added to measure the maximum mitochondrial cardiomyocyte maturation was improved within the SOTRs with respiration. Our results indicated that the maximum respiration

COMMUNICATIONS BIOLOGY | (2020) 3:122 | https://doi.org/10.1038/s42003-020-0853-0 | www.nature.com/commsbio 5 ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-0853-0

0 ReW vs 2 ReWs a Individuals factor map (PCA) b Spearman Correlation c 2

1ReW_2 1 1

0 2 ReWs_2 1ReW_1 0.99 2 ReWs_1 -1 0.98 -2 0ReW_1 0.97 Dim 2 (27.75%) 0.96

0.95 0ReW_2 0R 1 0 1 2 2 ReW R R R R eW eW_ e e e W Ws Ws _ _ _1 2 1 2 _ _ 2 Dim 1 (34.30%) 1 2 2 0R 1ReW 0 1R Re R R e e eW_ eW Ws_ Ws W _ _ _ _ 2 1 1 d 2 1 -Log 10 (P-value) 2 0 2 4 6 8 101214161820Total: 350

Response to unfolded protein GO:0006986 Count: 26

Muscle structure development GO:0061061 Count: 43

Extracellular structure organization GO:0043062 Count: 26

Actin filament-based process GO:0030029 Count: 35

Regulation of MAPK cascade GO:0043408 Count: 35

Tissue morphogenesis GO:0048729 Count: 30

Muscle system process GO:0003012 Count: 25

Vasculature development GO:0001944 Count: 34

Response to hypoxia GO:0001666 Count: 20

β -MHC / DAPI 䠅 e f **p=0.0053 2 β-MHC 䠄 1.5 nce e 1

0.5

0 0ReW 2ReWs Relative fluoresc 1ReW 0ReW 1ReW 2ReWs

g MYH7 ACTN2 DES MYL2

5 5 n 4 4 p= *p=0.0193 ** 0.0012 on p= **p=0.0043 i * 0.01826 sion ssio s

4 p= 4 s * 0.0423 s 3 3 pre re

3 3 ex xp 2 2

2 2 ene

ve g 1 1 1 1 i tive gene expre tive gene e a l elat a e 0 0 R 0 0 R Relative gene expression 0 ReW 1 ReW 2 ReWs Rel 0ReW 1ReW 2ReWs 0 ReW 1 ReW 2 ReWs 0 ReW 1 ReW 2 ReWs

MYL3 PLN MB ADRB1

3 4 n 4 8 p= ***p= o **p=0.0016 *** 0.0006 **p=0.0048 0.0002 i ion s s es ession es 3 r 3 6 2 r pr xp x e e e

e exp 4 2 2 e n n n

1 ge ge ege

e 2 1 v 1 i ve tive gene expression i tiv a at a l lat e Rel 0 0 Rel 0

Re 0 R 0ReW 1ReW 2ReWs 0ReW 1ReW 2 ReWs 0 ReW 1 ReW 2 ReWs 0 ReW 1 ReW 2 ReWs rate of groups with two ReWs was significantly higher than that SR ryanodine channels3. Although all groups responded to the observed in groups with zero or one ReW (Fig. 5a, b; 2 ReWs vs. 1 addition of 5 mM caffeine (Fig. 5c, d), SOTRs with two ReWs ReW: p = 0.0075; 2 ReWs vs. 0 ReW: p = 0.001), indicating that a exhibited significantly higher intensity changes than the groups higher number of ReWs led to increased mitochondrial activity. with zero ReW, indicating a higher maturation level in Ca2+- Because human cardiomyocytes rely on Ca2+ influx from the handling properties (p = 0.0237). Additionally, we recorded the sarcoplasmic reticulum (SR) for contraction, we investigated the contractile force of SOTRs using a mechanical tester (Fig. 5e, f SR function by applying caffeine to SOTRs in order to open the and Supplementary Fig. 14), which revealed that SOTRs with

6 COMMUNICATIONS BIOLOGY | (2020) 3:122 | https://doi.org/10.1038/s42003-020-0853-0 | www.nature.com/commsbio COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-0853-0 ARTICLE

Fig. 3 ReWs are associated with the upregulation of cardiac-specific gene expression. a PCA of SOTRs with or without ReWs based on RNA-sequencing data. b Heatmap showing hierarchical clustering of the correlation matrix resulting from comparison of expression values for each sample. Correlations calculated using Spearman’s correlation. c Heatmap showing the relative expression levels (z-score) of 349 genes from SOTRs with ReWs (1 and 2 ReWs) as compared with SOTRs with zero ReW. d GO categories significantly enriched for genes upregulated in SOTRs with two ReWs as compared with SOTRs with zero ReW. Genes with fold changes >1.5 and a P < 0.05 were analyzed. The enriched terms are listed. e Representative confocal images of SOTRs with zero, one, or two ReWs on day 14. Cardiomyocytes were stained with anti-β-MHC (red) and DAPI (blue). Scale bar represents 50 µm. f Quantification of β- MHC levels in SOTRs on day 14. The total integrated fluorescence of β-MHC acquired from confocal imaging was normalized to that of SOTRs with zero ReW (mean ± s.d.; n = 4 biologically independent samples). **P < 0.01 (ANOVA). g The expression of cardiac-specific genes (qPCR) in SOTRs 14 days after cell seeding (mean ± s.d.; n = 4 biologically independent samples). *P < 0.05; **P < 0.01; ***P < 0.001 (ANOVA).

either zero or two ReWs generated active forces that increased cardiac cells51. Further optimizations are needed to scale down with the applied stretching force in a Frank–Starling-like the PDMS mold and reduce the plated cell number. Additionally, fashion43. Moreover, the group with two ReWs generated forces because force generation is the primary function of cardiomyo- with larger amplitude than that in the group with zero ReW (2 cytes, force quantifications have been recently utilized to assess ReWs: 0.54 ± 0.15 mN mm−2; 0 ReW: 0.23 ± 0.12 mN mm−2). drug-related effects51,52; therefore, future mold design for SOTR formation might include a force transducer53, such as an elastic Discussion silicone pillar, capable of offering multiple parameters for more Excitation–contraction coupling is important for cardiomyocyte relevant assessment of drug response to pharmacologic agents. development and function44, and beating rate might play an Furthermore, since the electrical pacing could be controlled with important role in cardiac maturation. The human fetal heartbeat different loading levels (by gradually increasing or decreasing increases linearly during embryonic development from ~1 to ~3 frequency, or by tuning on/off the region of stimulation over Hz within 9 weeks in utero45,46. hiPSC-derived cardiomyocytes time)14,51, it would be useful to develop SOTRs with controllable functionally and morphologically resemble fetal cardiomyocytes; stimulation. SOTRs with controllable silicone pillar could be used however, the average beating rate of hiPSC-derived cardiomyo- to regulate the frequency of ReWs in real time. cytes is much lower than that of human fetal cardiomyocytes. It remains an open question as to what degree of cardio- Previous studies promoted cardiomyocyte maturation by myocyte maturation is necessary for regenerative applica- increasing their beating rates; however, the high beat rates of tions54, such as injection into or engraftment50 onto an hiPSC-derived cardiomyocytes could only be achieved by apply- infarcted heart. The various maturation levels of SOTRs under ing high-frequency electrical stimulation3,14,16, making the sus- different ReW training levels might offer preconditioned car- tainable application of such stimuli technically challenging. diomyocytes for delivery into an infarcted myocardium, thereby Moreover, electrical stimulation may cause potential side effects, allowing studies on how cardiomyocyte maturation affects including pH shift, ROS generation, electrolysis20, and cell therapeutic efficacy. damage18. The beating frequency and sarcomere length of ReW-trained In the present study, we created SOTRs with spontaneously cardiomyocytes did not increase with culture time (>89 days) generated ReWs capable of sustainably causing cardiomyocytes to longer than 14 days (Supplementary Fig. 6). This indicated that beat at high frequencies (up to 4 Hz) in the absence of external the rapid pacing maturation process could be limited within a stimulation. We found that cardiomyocytes with ReWs demon- specific time window, similar to the findings of a previous strated dramatically improved structural maturation, enhanced report14. Moreover, because the frequency of the ReWs within the cardiac gene expression, and Ca2+-handling properties in a SOTRs can be adjusted by changing the SOTR diameter, and frequency-dependent manner. Because this activity might also given that the ReWs could be sustained for >89 days, it might be cause hypoxia in SOTRs, in the future, an enhanced oxygen possible to create a pacemaker tissue with an adjustable beat rate supply, such as that afforded by dynamic culture5, can be utilized for use as an in vitro model for drug assessment or potentially for during SOTR culture. Additionally, although the cardiomyocytes in vivo heart pacing. trained by ReWs remained less mature than those generated by In conclusion, we found that ReWs could be spontaneously state-of-the-art methods11,14,47, the maturation level in ReWs generated and maintained within a SOTR comprising hiPSC- group could be further affected by changing the stimulation derived cardiomyocytes, and that the ReWs were able to make the window/duration and improving the beating frequency. cardiomyocytes beat at a high frequency comparable to that SOTRs were formed through a one-step, gel-free method uti- found in utero during embryonic development. Moreover, the lizing a polydimethylsiloxane (PDMS) mold and commercially ReWs are associated with structural and functional maturation of available Petri dishes. SOTR formation required <2 days, which is the cardiomyocytes, thereby offering a supplementary approach a much shorter period than the previously reported tissue ring to electrical stimulation-based maturation of electrically active formation time (~7–14 days)22,48,49, owing to multiple factors, cell types. including the use of gel-free medium, the low attachment surface of the culture dish, and the ability of the cardiomyocytes to aggregate into a 3D construct. This methodology could be an Methods important complement to current widely used methods3,20,22,26, Differentiation and culture of hiPSC-derived cardiomyocytes. GCaMP3- positive hiPSCs (253G1) were cultured and differentiated according to previously and the rapid formation of the tissue ring without using any 55,56 fi published methods . All experiments involving the use of hiPSCs were per- exogenous ECM, such as collagen or brin, might provide new formed following Kyoto University and Osaka University guidelines. After 30 to insights into future mass production of mature tissues for drug 50 days of differentiation, cardiomyocyte colonies floating in the medium were screening and regenerative applications50 collected and dissociated into a single-cell suspension by stirring for between 1 and −1 Given the large cell number required for SOTR formation in 2 h in protease solution [0.1% collagenase type I, 0.25% trypsin, 1 U mL DNase I, 5 116 mM NaCl, 20 mM HEPES, 12.5 mM NaH2PO4, 5.6 mM glucose, 5.4 mM KCl, one well (4 × 10 cells), it remains difficult to achieve a high- 56 and 0.8 mM MgSO4 (pH 7.35)] . The cardiomyocyte purity was characterized throughput assessment that is comparable to that of a previous with flow cytometry and the cardiomyocytes with high purity (>85%) were used for study where 200 tissue constructs were created with per million following experiments.

COMMUNICATIONS BIOLOGY | (2020) 3:122 | https://doi.org/10.1038/s42003-020-0853-0 | www.nature.com/commsbio 7 ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-0853-0

a 0ReW 1ReW 2 ReWs DAPI / T2 Tn

b 90 90 90 120 60 120 60 120 60

150 30 150 30 150 30

180 0 180 0 180 0

150 30 150 30 150 30

120 60 120 60 120 60 90 90 90 c 0ReW 1ReW 2 ReWs DAPI / in actin - α

d 2 **p = 0.001 1.9 1.8 *p=0.012 0ReW 1.7 th (μm) 1.6 1.5 eleng 1.4 1.3 arcomer S 1.2 1.1 1 0ReW 1ReW 2ReWs

1ReW 2 ReWs

Zline I-band

A-band

Fig. 4 ReWs are associated with cardiomyocyte alignment. a Representative confocal images of SOTRs with zero, one, or two ReWs. cardiomyocytes were stained with anti-TnT2 (red) and DAPI (blue), revealing cardiomyocyte alignment in ReW samples at day 14. Arrows mark the ring orientation, and the area delimited by a green dashed square is zoomed-in in each confocal image. Scale bar represents 50 µm. b Angular graph based on Fourier component analysis of TnT2 orientation distribution. c Representative confocal images of stained SOTRs with different ReWs (red: α-actinin; blue: DAPI). Arrows mark the ring orientation at day 14. Scale bar represents 30 µm. d Sarcomere lengths of cardiomyocytes within SOTRs on day 14 [mean ± s.d.; 0 ReW (n = 5); 1 ReW (n = 5); 2 ReWs (n = 4 biologically independent samples)]. *P < 0.05; **P < 0.01 (ANOVA). e TEM analysis of SOTRs on day 14. Scale bar represents 1 µm.

8 COMMUNICATIONS BIOLOGY | (2020) 3:122 | https://doi.org/10.1038/s42003-020-0853-0 | www.nature.com/commsbio COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-0853-0 ARTICLE

a b 100 80 0 ReW ROTE/ANTI **p = 0.001 1 ReW 70 ) p = -1 80 2 ReWs ** 0.0075 FCCP 60

Oligo ) 1 60 - 50 n i

lm 40

40 o

m 30 (p

OCAR(pmol min 20 20 10 0 Maximal respiration capacity capacity respiration Maximal 0 6 13 20 27 33 40 47 56 65 73 82 89 96 103 109 0 Time (min) 0 ReW 1 ReW 2 ReWs c 0 ReW Caffeine 5mM

0.2 F/F0 d

0.4 Before caffeine addition After caffeine addition *p = 0.0237

0 0.3 Caffeine 5mM 1 ReW .2 F/F 0 0.2 peak intensity peak intensity Before caffeine addition After caffeine addition 0.1 Caffeine-induced change of

0 2 ReWs Caffeine 5mM 0 ReW 1 ReW 2 ReWs 0.2 F/F0 0.2

Before caffeine addition After caffeine addition

2 ReWs f 0.8 e 0 ReW *p = 0.044

e 0.7 Stretch 10% 0.6 Stretch 20% )

-2 0.5 Stretch 30% 0.4 contraction forc contraction (mN mm

m 0.3

0.2 -2 Maximu 0.1

0 0 ReW 2 ReWs 1 mN mm

0. 0.4 s

Fig. 5 ReWs are associated with enhanced Ca2+-handling properties. a Results of Oxygen Consumption Rate (OCR) assays. Oligo, oligomycin; FCCP, carbonyl cyanide-4-(triﬂuoromethoxy) phenylhydrazone; ROTE/ANTI, rotenone and antimycin A. b Quantiﬁcation of the maximal respiration capacity (mean ± s.d.; n = 4 biologically independent samples). **P < 0.01 (ANOVA). c Representative traces of Ca2+ transient following administration of 5 mM caffeine to cardiomyocytes in SOTRs with zero ReW, one ReW, and two ReWs at day 7. SOTRs were immersed in room temperature medium to stop ReWs on day 6. d Caffeine-induced change in peak intensity in different groups (mean ± s.d.; n = 6 biologically independent samples). *P < 0.05 (ANOVA). e Force traces of SOTRs during progressive stretching. f Maximum contractile force of SOTRs (mean ± s.d.; n = 3 biologically independent samples). *P < 0.05 (one-tailed Student’s t test).

Device manufacture. PDMS (SYLGARD 184; Dow Corning, Midland, MI, USA) number was fixed to 4 × 105 or to a cell number that was proportional to the blocks were punched using a tissue puncher (Thermo Fisher Scientific, Waltham, diameter. After plating, cardiomyocytes settled in the wells, aggregated, and con- MA, USA) to an inner diameter of 8 mm. After removal of the redundant central gregated around the central pillar to form densely packed tissue rings within 2 days. pillar, the remaining PDMS wells were collected, and PDMS wells and pillars with The medium was changed to serum-free medium (culture medium without FBS) different diameters were centrally aligned and attached to the bottom of 24-well from day 2. After that, fresh medium was changed every 4 days. Before changing ultra-low attachment plates (Corning, Corning, NY, USA) as shown in Supple- the medium, the fresh medium was preheated to 37 °C. The dishes were placed on mentary Fig. 1c. For micro-electrode array (MEA; Multi Channel Systems, Reu- top of a preheated metal block during medium change. The medium was pipetted tlingen, Germany) recording, the PDMS well and pillars were mounted to the MEA gently and slowly into the wells. For dynamic culture, the dishes were mounted on surface using silicone glue (Shin-Etsu Chemical, Tokyo, Japan) to fix the well and a rotary shaker (NA-M301, Nissin, Japan), rotating at a speed of 26 r.p.m., in an pillar to the bottom of the well. After drying and treatment with ultraviolet light for incubator for 2 weeks. All the data in this study were collected under static culture 30 min, the PDMS wells were ready for cardiomyocyte culture. condition for 14 days, unless specified otherwise.

SOTR generation. Before cell seeding, single cardiomyocytes were filtered using a Optical mapping and measurement. Progression of tissue assembly was observed 40-µm cell strainer (BD Falcon; Becton Dickenson, Franklin Lakes, NJ, USA) and using a fluorescence microscope (Olympus IX71; Olympus, Tokyo, Japan) equip- resuspended at a density of 2 × 106 cells mL−1 in the culture medium containing ped with a charge-coupled device (CCD) camera (Exiblue; Qimaging, Surrey, BC, 40% high glucose Dulbecco’s modified Eagle’s medium (DMEM) (Sigma-Aldrich), Canada; or DP74; Olympus). GCaMP3 was excited from 450 to 490 nm, and 40% IMDM (Iscove’s modified Dulbecco’s medium; Sigma-Aldrich), 20% fetal fluorescence images of GCaMP3-positive cardiomyocytes were recorded with 8 × 8 bovine serum (FBS; Gibco, USA), 1% minimum essential medium non-essential binning of CCD pixels at 30 frames s−1. Images were processed to obtain data amino acid solution (Sigma-Aldrich), and 0.1% penicillin–streptomycin (Gibco), using ImageJ (NIH, Bethesda, MD, USA) and MATLAB (R2014b; MathWorks, and 0.5% L-glutamine (Sigma-Aldrich). Then, 4 × 105 cells were plated in each Natick, MA, USA) using a customized program. Activation-time mapping was PDMS culture well with a 3-mm pillar. For wells with other diameters, the cell performed using a custom plug-in in ImageJ, as previously described57. Briefly, a

COMMUNICATIONS BIOLOGY | (2020) 3:122 | https://doi.org/10.1038/s42003-020-0853-0 | www.nature.com/commsbio 9 ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-0853-0

Gaussian spatial convolution was applied with a 2 pixel radius to each frame. 30 min, incubated with anti-TnT2 antibodies (mouse monoclonal IgG, 1:200; Santa Subsequently, the time-series data were temporal low pass filtered for noise Cruz Biotechnology: SC-20025) or isotype-matched antibodies (BD Phosphoflow: reduction, the upstrokes of cell activation were detected, and a linear activation 557782) at 37 °C for 30 min, washed with D-PBS, and then incubated with Alexa map was calculated between subsequent activations independently for each pixel. Fluor 488 anti-mouse IgG (1:500; Jackson ImmnoResearch: 715-546-150). Cells were then washed twice with D-PBS and analyzed using a FACS Canto II flow cytometer (BD Biosciences, USA) and the FlowJo software (Treestar Inc., USA). Voltage-sensitive dye staining. The voltage dye staining was performed Data shown are representative of four independent experiments. according to the manufacturer’s protocol (FluoVolt™, Thermo Fisher Scientific). Five microliters of FluoVoltTM dye was diluted in 50 µL PowerLoad and then suspended in 5 mL of high glucose DMEM (Sigma-Aldrich) preheated to 37 °C. Contractility analysis. SOTR contractility was measured using a micron-scale Then, 200 µL of the dye solution was applied to the well containing SOTRs for 15 mechanical-testing system (MicroSquisher; CellScale Biomaterials Testing, min. After removing the dye solution, the tissues were washed with DMEM twice. Waterloo, ON, Canada). ReWs were stopped by exchanging the medium at room temperature 1 day before the measurement, and the SOTR was removed from the + + pillar and cut into a strip. Then, the strip was fixed on a stage and immersed in the Transient and electrophysiological Ca2 characterization.Ca2 transience was culture medium warmed at 37 °C. A cantilever beam with a diameter of 0.30 mm recorded by fluorescence imaging of GCaMP3-positive cardiomyocytes with the was pressed onto the SOTR (Supplementary Fig. 14). The beam were lower to same experimental setup as that described for optical mapping and measurement. stretch the SOTR strip into different lengths. At each length, the beam was held for To characterize the spontaneous beating of SOTRs, ReWs were removed from the 50 s and the force was calculated by cantilever beam deflection in response to SOTRs 1 day before recording by emerging the SOTRs into the medium at room differential displacement. temperature for 1 min. The 5 mM caffeine (Wako Pure Chemical Industries Ltd. Osaka, Japan) was added directly to the chamber containing the SOTRs during imaging, as described in the figure legends. Quantitative polymerase chain reaction. Total RNA in SOTRs was harvested Extracellular recording of field potentials (FPs) was performed using an MEA using Trizol reagent (Life Technologies) according to the manufacturer’s instructions, data-acquisition system (USB-ME64; Multi Channel Systems). Signals were and RNA concentration was determined using a NanoDrop1000 spectrophotometer recorded from days 5 and 6 after cardiomyocyte seeding, and data were collected (Thermo Fisher Scientific). Complementary DNA (cDNA) was synthesized with a and processed using MC_Rack (Multi Channel Systems) or LabChart First-strand synthesis kit (TaKaRa, Shiga, Japan) and analyzed by qPCR using SYBR (ADInstruments, Dunedin, New Zealand). The amplitude, QT interval, and beat Green PCR master mix (Life Technologies) and the qBiomarker validation PCR array rate were determined by analyzing the wave form of the FP. (IPHS-102A; Qiagen, Hilden, Germany) in a 96-well format according to the manufacturer’s instructions. Cycling conditions were set as follows: initial denaturation at 95 °C for 10 min, followed by 40 cycles at 95 °C for 15 s and 60 °C for 70 s. Reactions Histology. SOTRs were rinsed three times with phosphate-buffered saline (PBS), were performed in a StepOnePlus real-time PCR system (Life Technologies). Gene fixed in 4% paraformaldehyde (PFA) in PBS for 30 min, and embedded in paraffin. expression was determined using the 2−ΔΔCt method and relative to glyceraldehyde 3- Thin sections were sliced and stained with HE (Muto Chemical Corporation, phosphate dehydrogenase expression. Heatmaps were created by the “pheatmap” Tokyo, Japan), and observation was carried out using a CKX41 microscope package (https://cran.r-project.org/web/packages/pheatmap/index.html), and clus- (Olympus). tering order was produced using the Ward.D clustering algorithm (https://stat.ethz. ch/R-manual/R-devel/library/stats/html/hclust.html). Immunostaining and imaging. SOTRs were fixed in 4% PFA at room temperature for 30 min, permeabilized with 0.5% (v/v) Triton X-100 in Dulbecco’s (D)-PBS at RNA quantification and qualification for RNA-sequencing. RNA degradation room temperature for 1 h, and incubated in blocking solution [0.1% (v/v) Tween- and contamination was monitored on 1% agarose gels. RNA purity was determined 20, 5% (v/v) normal donkey serum, 3% (v/v) bovine serum albumin, and 5% (v/v) using a NanoPhotometer spectrophotometer (Implen, Westlake Village, CA, USA). normal goat serum in D-PBS] at 4 °C for 16 h. SOTRs were then incubated with the RNA concentration was measured using a Qubit RNA assay kit in a Qubit 2.0 primary antibodies anti-troponin T2 (TnT2; mouse monoclonal IgG; 1:200; SC- fluorometer (Life Technologies). RNA integrity was assessed using the RNA Nano 20025; Santa Cruz Biotechnology, Dallas, TX, USA), anti-α-actinin (mouse 6000 assay kit and the Agilent Bioanalyzer 2100 system (Agilent Technologies). monoclonal IgG, 1:1000; A7811; Sigma-Aldrich), or anti-β-MHC (mouse MYH7 monoclonal IgM, 1:100; SC-53089; Santa Cruz Biotechnology) at 4 °C overnight. SOTRs were then rinsed with PBS and incubated with secondary antibodies diluted Library preparation for RNA-sequencing. A total of 1.5 μg RNA per sample was 1:300 in blocking buffer (Alexa Fluor 594 anti-mouse IgG; 715-586-150; and used as input material for RNA sample preparations. Adult normal heart RNA was DyLight-594 anti-mouse IgM; 715-516-020; Jackson ImmnoResearch, West Grove, purchased from BioChain (USA). Sequencing libraries were generated using a PA, USA) at room temperature for 1 h. 4′-6-Diamidino-2-phenylindole (DAPI; NEBNext UltraTM RNA library prep kit Illumina (New England Biolabs, Ipswich, 300 nM; Wako Pure Chemical Industries, Ltd.) was used to counterstain nuclei at MA, USA) according to the manufacturer’s instructions, and index codes were room temperature for 30 min, after which images were captured using a confocal added to attribute sequences to each sample. Briefly, mRNA was purified from total microscope (FV1200; Olympus). cardiomyocyte orientation within SOTRs was RNA using poly-T oligo-attached magnetic beads, and fragmentation was per- determined using the Fourier component analysis plug-in “Directionality” in formed using divalent cations under an elevated temperature in NEBNext First- ImageJ46 (NIH) that calculated the orientation distribution for the red color Strand Synthesis reaction buffer (5×; New England Biolabs). First-strand cDNA channel. Fluorescence quantification of β-MHC was performed by calculating the was synthesized using a random hexamer primer and M-MuLV reverse tran- gray value averaged over the area in all sample groups, with all the fluorescence scriptase (RNase H−; New England Biolabs). Second-strand cDNA synthesis was values were normalized to that of zero ReW group. subsequently performed using DNA polymerase I and RNase H. Remaining overhangs were converted into blunt ends via exonuclease/polymerase activities. After adenylation of the 3′ ends of the DNA fragments, NEBNext Adaptor (New Transmission electron microscopy (TEM). SOTRs were washed with 0.1 M England Biolabs) with a hairpin loop structure was ligated to prepare for hybri- Sorenson’s buffer (pH 7.4) twice and fixed with 2.5% glutaraldehyde (Sigma- dization. To select cDNA fragments of the correct length, the library fragments Aldrich) in 0.1 M Sorenson’s buffer (pH 7.4), after which samples were post-fixed were purified using the AMPure XP system (Beckman Coulter, Beverly, MA, USA), with 1% OsO in Sorrenson’s buffer. The samples were then embedded, sliced, and 4 after which 3 µL USER enzyme (New England Biolabs) was incubated with size- stained with lead citrate and examined under a JEOL1010 transmission electron selected, adaptor-ligated cDNA at 37 °C for 15 min, followed by 5 min at 95 °C. microscope (JEOL Ltd., Tokyo, Japan). PCR was performed using a Phusion high-fidelity DNA polymerase, universal PCR primers, and an Index (X) primer (New England Biolabs). Products were purified Mitochondrial respiration assay. Mitochondrial function was analyzed using a (AMPure XP; Beckman Coulter), and library quality was assessed using the Agilent Seahorse XF96 extracellular flux analyzer (Agilent Technologies, Carlsbad, CA, Bioanalyzer 2100 system (Agilent Technologies). USA). After a 14-day culture, SOTRs were dissociated into a single-cell suspension by stirring for 30 min in protease solution, followed by cell seeding onto a cell Clustering and sequencing. Clustering of the Index-coded samples was performed culture microplate (Agilent Technologies) at a density of 20,000 cells per XF96 well. on a cBot cluster generation system using a HiSeq 4000 PE cluster kit (Illumia, San OCR assays were performed 3 days after seeding, and the culture medium was Diego, CA, USA) according to the manufacturer’s instructions. After cluster gen- changed to the base medium (Seahorse XF assay media supplemented with 1 mM eration, the library preparations were sequenced on an Illumina Hiseq 4000 plat- sodium pyruvate; Life Technologies, Carlsbad, CA, USA). Substrates and inhibitors form (Illumina), and 150-bp paired-end reads were generated. were injected during measurements at a final concentration of 3.5 μM 4-(tri- After quality control, the paired-end reads were aligned to the hg19 human fluoromethoxy) phenylhydrazone (FCCP; Seahorse Bioscience, Billerica, MA, reference genome (UCSC; https://genome.ucsc.edu/) using TopHat (v2.0.12; USA), 1 μM oligomycin, 0.5 μM antimycin, and 0.5 μM rotenone for the https://ccb.jhu.edu/software/tophat/index.shtml)58. We quantified the relative MitoStress assay. expression level as log2(FPKM + 1) to perform PCA analysis and hierarchical clustering. Uniquely mapped reads counted by HTSeq59 were used to conduct Flow cytometry. Before being used for generating SOTRs, hiPSC-derived cardio- differential expression analysis in DEseq260. A total of 349 genes (P < 0.05, fold myocytes were fixed in 4% PFA at room temperature for 30 min, permeabilized change >1.5) were selected for the GO analysis. We plotted heatmaps using the with 0.5% v/v Triton X-100 in Dulbecco’s (D)-PBS at room temperature for “pheatmap” package in R (https://cran.r-project.org/web/packages/pheatmap/

10 COMMUNICATIONS BIOLOGY | (2020) 3:122 | https://doi.org/10.1038/s42003-020-0853-0 | www.nature.com/commsbio COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-0853-0 ARTICLE index.html) and performed GO enrichment analysis using Metascape61 (http:// 19. Hirt, M. N. et al. Functional improvement and maturation of rat and human metascape.org/). engineered heart tissue by chronic electrical stimulation. J. Mol. Cell. Cardiol. 74, 151–161 (2014). Statistics and reproducibility. All quantitative data are presented as the mean ± 20. Godier-Furnémont, A. F. et al. Physiologic force-frequency response in standard deviation of the mean (mean ± s.d.). The differences between experi- engineered heart muscle by electromechanical stimulation. Biomaterials 60, – mental groups were analyzed by one-tailed Student’s t test (between two groups) or 82 91 (2015). ANOVA (one-way analysis of variance), followed by Tukey’s post hoc test (among 21. van Meer, B. J., Tertoolen, L. G. & Mummery, C. L. Measuring physiological three or more groups). A P < 0.05 was considered statistically significant. responses of human pluripotent stem cell derived cardiomyocytes to drugs and disease. Stem Cells 34, 2008–2015 (2016). 22. Zimmermann, W.-H. et al. Tissue engineering of a differentiated cardiac Reporting summary . Further information on research design is available in muscle construct. Circ. Res. 90, 223–230 (2002). the Nature Research Reporting Summary linked to this article. 23. Dhein, S. et al. Mechanical control of cell biology. Effects of cyclic mechanical stretch on cardiomyocyte cellular organization. Prog. Biophys. Mol. Biol. 115, Data availability 93–102 (2014). The RNA-sequencing data in this paper could be visited under the accession number 24. Li, Y. et al. Chinese-noodle-inspired muscle myofiber fabrication. Adv. Funct. GEO: GSE140466. The other datasets generated during the current study are available Mater. 25, 5999–6008 (2015). from the corresponding author on reasonable request. 25. Tulloch, N. L. et al. Growth of engineered human myocardium with mechanical loading and vascular coculture. Circ. Res. 109,47–59 (2011). 26. Mihic, A. et al. The effect of cyclic stretch on maturation and 3D tissue Received: 10 June 2019; Accepted: 19 February 2020; formation of human embryonic stem cell-derived cardiomyocytes. Biomaterials 35, 2798–2808 (2014). 27. Shaheen, N. et al. Human induced pluripotent stem cell-derived cardiac cell sheets expressing genetically encoded voltage indicator for pharmacological and arrhythmia studies. Stem Cell Rep. 10, 1879–1894 (2018). 28. Kadota, S. et al. Development of a reentrant arrhythmia model in human References pluripotent stem cell-derived cardiac cell sheets. Eur. Heart J. 34, 1147–1156 1. Karakikes, I., Ameen, M., Termglinchan, V. & Wu, J. C. Human induced (2012). pluripotent stem cell–derived cardiomyocytes insights into molecular, cellular, 29. González, H., Nagai, Y., Bub, G., Glass, L. & Shrier, A. Reentrant waves in a + and functional phenotypes. Circ. Res. 117,80–88 (2015). ring of embryonic chick ventricular cells imaged with a Ca2 sensitive dye. 2. Dolnikov, K. et al. Functional properties of human embryonic stem BioSystems 71,71–80 (2003). cell-derived cardiomyocytes: intracellular Ca2+ handling and the role 30. Kawatou, M. et al. Modelling Torsade de Pointes arrhythmias in vitro in 3D of sarcoplasmic reticulum in the contraction. Stem Cells 24, 236–245 human iPS cell-engineered heart tissue. Nat. Commun. 8, 1078 (2017). (2006). 31. Goldfracht, I. et al. Engineered heart tissue models from hiPSC-derived 3. Nunes, S. S. et al. Biowire: a platform for maturation of human pluripotent cardiomyocytes and cardiac ECM for disease modeling and drug testing stem cell-derived cardiomyocytes. Nat. Methods 10, 781–787 (2013). applications. Acta Biomater. 92, 145–159 (2019). 4. Davis, R. P. et al. Cardiomyocytes derived from pluripotent stem cells 32. Smeets, J. L., Allessie, M. A., Lammers, W. J., Bonke, F. I. & Hollen, J. The recapitulate electrophysiological characteristics of an overlap syndrome of wavelength of the cardiac impulse and reentrant arrhythmias in isolated rabbit cardiac sodium channel disease. Circulation 125, 3079–3091 (2012). atrium. The role of heart rate, autonomic transmitters, temperature, and 5. Jackman, C. P., Carlson, A. L. & Bursac, N. Dynamic culture yields engineered potassium. Circ. Res. 58,96–108 (1986). myocardium with near-adult functional output. Biomaterials 111,66–79 33. Izumi-Nakaseko, H. et al. Characterization of human iPS cell-derived (2016). cardiomyocyte sheets as a model to detect drug-induced conduction 6. Ulmer, B. M. et al. Contractile work contributes to maturation of energy disturbance. J. Toxicol. Sci. 42, 183–192 (2017). metabolism in hiPSC-derived cardiomyocytes. Stem Cell Rep. 10, 834–847 34. Rensma, P. L., Allessie, M. A., Lammers, W. J., Bonke, F. I. & Schalij, M. J. (2018). Length of excitation wave and susceptibility to reentrant atrial arrhythmias in 7. Shadrin, I. Y. et al. Cardiopatch platform enables maturation and scale-up of normal conscious dogs. Circ. Res. 62, 395–410 (1988). human pluripotent stem cell-derived engineered heart tissues. Nat. Commun. 35. Courtemanche, M., Keener, J. P. & Glass, L. A delay equation representation 8, 1825 (2017). of pulse circulation on a ring in excitable media. SIAM J. Appl. Math. 56, 8. Thavandiran, N. et al. Functional arrays of human pluripotent stem cell- 119–142 (1996). derived cardiac microtissues. bioRxiv 566059 (2019). 36. Courtemanche, M., Glass, L. & Keener, J. P. Instabilities of a propagating pulse 9. Gao, L. et al. Myocardial tissue engineering with cells derived from human- in a ring of excitable media. Phys. Rev. Lett. 70, 2182 (1993). induced pluripotent stem cells and a native-like, high-resolution, 3- 37. Shaw, R. M. & Rudy, Y. Ionic mechanisms of propagation in cardiac tissue: dimensionally printed scaffold. Circ. Res. 120, 1318–1325 (2017). roles of the sodium and L-type calcium currents during reduced excitability 10. Parikh Shan, S. et al. Thyroid and glucocorticoid hormones promote and decreased gap junction coupling. Circ. Res. 81, 727–741 (1997). functional T-tubule development in human-induced pluripotent stem cell- 38. Beauchamp, P. et al. Electrical propagation in synthetic ventricular myocyte derived cardiomyocytes. Circ. Res. 121, 1323–1330 (2017). strands from germline connexin43 knockout mice. Circ. Res. 95, 170–178 11. Herron, T. J. et al. Extracellular matrix–mediated maturation of human (2004). pluripotent stem cell-derived cardiac monolayer structure and 39. Arduini, D., Rizzo, G. & Romanini, C. Fetal Cardiac Function (Parthenon Pub. electrophysiological function. Circ. Arrhythm. Electrophysiol. 9, e003638 Group, New York, 1995). (2016). 40. Groenendyk, J., Agellon, L. B. & Michalak, M. Coping with endoplasmic 12. Brevet, A., Pinto, E., Peacock, J. & Stockdale, F. E. Myosin synthesis increased reticulum stress in the cardiovascular system. Annu. Rev. Physiol. 75,49–67 by electrical stimulation of skeletal muscle cell cultures. Science 193, (2013). 1152–1154 (1976). 41. Branco, M. A. et al. Transcriptomic analysis of 3D cardiac differentiation of 13. McDonough, P. M. & Glembotski, C. Induction of atrial natriuretic factor and human induced pluripotent stem cells reveals faster cardiomyocyte myosin light chain-2 gene expression in cultured ventricular myocytes by maturation compared to 2D culture. Sci. Rep. 9, 9229 (2019). electrical stimulation of contraction. J. Biol. Chem. 267, 11665–11668 (1992). 42. van der Velden, J., de Jong, J. W., Owen, V., Burton, P. & Stienen, G. 14. Ronaldson-Bouchard, K. et al. Advanced maturation of human cardiac tissue Effect of protein kinase A on calcium sensitivity of force and its sarcomere grown from pluripotent stem cells. Nature 556, 239–243 (2018). length dependence in human cardiomyocytes. Cardiovasc. Res. 46, 487–495 15. Chan, Y.-C. et al. Electrical stimulation promotes maturation of (2000). cardiomyocytes derived from human embryonic stem cells. J. Cardiovasc. 43. Asnes, C. F., Marquez, J. P., Elson, E. L. & Wakatsuki, T. Reconstitution of the Transl. Res. 6, 989–999 (2013). Frank–Starling mechanism in engineered heart tissues. Biophys. J. 91, 16. Eng, G. et al. Autonomous beating rate adaptation in human stem cell-derived 1800–1810 (2006). cardiomyocytes. Nat. Commun. 7, https://doi.org/10.1038/ncomms10312 44. Stoppel, W. L., Kaplan, D. L. & Black Iii, L. D. Electrical and mechanical (2016). stimulation of cardiac cells and tissue constructs. Adv. Drug Deliv. Rev. 96, 17. Tandon, N. et al. Electrical stimulation systems for cardiac tissue engineering. 135–155 (2016). Nat. Protoc. 4, 155–173 (2009). 45. Lindsley, D. B. Heart and brain potentials of human fetuses in utero. Am. J. 18. Merrill, D. R., Bikson, M. & Jefferys, J. G. R. Electrical stimulation of excitable Psychol. 55, 412–416 (1942). tissue: design of efficacious and safe protocols. J. Neurosci. Methods 141, 46. DuBose, T. J., Cunyus, J. A. & Johnson, L. F. Embryonic heart rate and age. J. 171–198 (2005). Diagn. Med. Sonogr. 6, 151–157 (1990).

COMMUNICATIONS BIOLOGY | (2020) 3:122 | https://doi.org/10.1038/s42003-020-0853-0 | www.nature.com/commsbio 11 ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-0853-0

47. Zhang, D. et al. Tissue-engineered cardiac patch for advanced functional Author contributions maturation of human ESC-derived cardiomyocytes. Biomaterials 34, J.L., L.L., C.T., and Y.S. conceived the project, and J.L., L.Y., I.M., C.Y., L.L., C.T., S.M., 5813–5820 (2013). and Y.S., designed the experiments. J.L. and J.Q. fabricated the device. Y.Z. and J.L. 48. Gwyther, T. A. et al. Engineered vascular tissue fabricated from aggregated performed electron microscopy and analyzed the images. Y.S. generated the GCaMP3- smooth muscle cells. Cells Tissues Organs 194,13–24 (2011). expressing hiPSC line. I.M. performed hiPSC culture and cardiomyocyte differentiation. 49. Guo, X.-M. et al. Creation of engineered cardiac tissue in vitro from mouse J.L., L.Y., X.Q., Y.H., M.H., and N.F. conducted the immunostaining experiments. L.Y., embryonic stem cells. Circulation 113, 2229–2237 (2006). J.D., Y.Z., and F.T. performed the qRT-PCR and RNA-sequencing experiments. J.L., 50. Weinberger, F. et al. Cardiac repair in guinea pigs with human engineered M.H., and J.Q. performed the electrophysiological characterization. L.Y., I.M., and J.L. heart tissue from induced pluripotent stem cells. Sci. Transl. Med. 8, performed flow cytometry. L.Z. and C.T. performed the simulation experiments. All 363ra148–363ra148 (2016). authors contributed to data analysis and interpretation. J.L., L.Z., Y.S., C.T., and L.L. 51. Boudou, T. et al. A microfabricated platform to measure and manipulate the wrote the manuscript. mechanics of engineered cardiac microtissues. Tissue Eng. Part A 18, 910–919 (2011). Competing interests 52. Agarwal, A., Goss, J. A., Cho, A., McCain, M. L. & Parker, K. K. Microfluidic The authors declare no competing non-financial interests but the following competing heart on a chip for higher throughput pharmacological studies. Lab a Chip. fi fi 13, 3599–3608 (2013). nancial interests: L.L. and J.L. led a provisional Japanese patent application (2016- 255258) based on the research presented here. I.M. is a shareholder of Myoridge Co. Ltd. 53. Marquez, J. P. et al. High-throughput measurements of hydrogel tissue ’ construct mechanics. Tissue Eng. Part C 15, 181–190 (2009). Y.S. s laboratory received funding from the TERUMO Company. 54. Laflamme, M. A. & Murry, C. E. Heart regeneration. Nature 473, 326–335 (2011). Additional information 55. Shiba, Y. et al. Human ES-cell-derived cardiomyocytes electrically couple and Supplementary information is available for this paper at https://doi.org/10.1038/s42003- suppress arrhythmias in injured hearts. Nature 489, 322–325 (2012). 020-0853-0. 56. Minami, I. et al. A small molecule that promotes cardiac differentiation of human pluripotent stem cells under defined, cytokine- and xeno-free Correspondence and requests for materials should be addressed to Y.S., C.T. or L.L. conditions. Cell Rep. 2, 1448–1460 (2012). 57. Isomura, A., Hörning, M., Agladze, K. & Yoshikawa, K. Eliminating spiral Reprints and permission information is available at http://www.nature.com/reprints waves pinned to an anatomical obstacle in cardiac myocytes by high- frequency stimuli. Phys. Rev. E 78, 066216 (2008). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in 58. Trapnell, C., Pachter, L. & Salzberg, S. L. TopHat: discovering splice junctions published maps and institutional affiliations. with RNA-Seq. Bioinformatics 25, 1105–1111 (2009). 59. Anders, S., Pyl, P. T. & Huber, W. HTSeq—a Python framework to work with high-throughput sequencing data. Bioinformatics 31, 166–169 (2015). 60. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 Open Access This article is licensed under a Creative Commons (2014). Attribution 4.0 International License, which permits use, sharing, 61. Tripathi, S. et al. Meta-and orthogonal integration of influenza “OMICs” data defines a role for UBR4 in virus budding. Cell Host Microbe 18, 723–735 adaptation, distribution and reproduction in any medium or format, as long as you give (2015). appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the Acknowledgements article’s Creative Commons license and your intended use is not permitted by statutory Funding was provided by the Japan Society for the Promotion of Science (JSPS): Grant-in- regulation or exceeds the permitted use, you will need to obtain permission directly from Aid for Challenging Exploratory Research (B) (15K13911, to L.L.), Grants-in-Aid for Sci- the copyright holder. To view a copy of this license, visit http://creativecommons.org/ fi enti c Research (C) (19K12801, to J.L.), Grants-in-Aid for Young Scientists (B) (17K14624, licenses/by/4.0/. to J.L.). Funding was also provided by the Japan Agency for Medical Research and Development (AMED). This work was also supported by the Ministry of Science and Technology of China (2015CB910300, to C.T. and 2018YFA0800504 to Y.Z.). © The Author(s) 2020

12 COMMUNICATIONS BIOLOGY | (2020) 3:122 | https://doi.org/10.1038/s42003-020-0853-0 | www.nature.com/commsbio